Ben McLaughlin, Janet Peterson, & Ming Ye


Solute Transport: An Application of Model Reduction


Being able to accurately and reliably model subsurface systems is integral to our ability to understand, monitor, and conserve groundwater, one of our most vital resources. Many analysis tools used in contaminant transport modeling, such as parameter estimation and uncertainty quantification, have an extremely high computational cost because of the high number of realizations of the forward model that are required. In previous research we have put forth the technique of reduced order modeling (ROM) via proper orthogonal decomposition (POD) as a viable means to solve diffusion-dominated linear reactive transport systems, demonstrating the method in the one-dimensional case, but did not present any data regarding the computational savings. Here we demonstrate a two-dimensional example, and the computational cost-savings of the reduced model. We will describe the processes of reactive transport, outline the method of model reduction, and demonstrate the ability of ROM to improve simulation time while maintaining accuracy. As our work continues, we turn our attention to some of the methods we are researching for treating more realistic reactive transport scenarios.